Optimal Differentiation of Tissue Types Using Combined Mid and Near Infrared Spectroscopy Mugdha V. Padalkar, M.S. 1, Cushla M. McGoverin, Ph.D. 1, Uday P. Palukuru, M.S. 1, Nicholas J. Caccese 1, Padraig B. Glenn 1, Scott Barbash, M.D. 2, Eric Kropf, M.D. 2, Nancy Pleshko, Ph.D. 1. 1 Department of Bioengineering, Temple University, Philadelphia, PA, USA, 2 Department of Orthopaedic Surgery and Sports Medicine, Temple University, Philadelphia, PA, USA. Disclosures: M.V. Padalkar: None. C.M. McGoverin: None. U.P. Palukuru: None. N.J. Caccese: None. P.B. Glenn: None. S. Barbash: None. E. Kropf: None. N. Pleshko: None. Introduction: Despite the number of anterior cruciate ligament reconstructions performed every year, the process of ligamentization, transformation of a tendon graft to a healthy functional ligament [1], is poorly understood. Fourier transform infrared (FT-IR) spectroscopy is a technique sensitive to molecular structure and composition changes in tissues. FT-IR fiber optic probes combined with arthroscopy could prove to be an important tool where nondestructive tissue assessment is required, such as assessment of graft composition during the ligamentization process. The mid-ir spectral absorbances from connective tissues are well understood, but mid-ir radiation has limited penetration, through only ~10 microns of the tissue. In contrast, near infrared (NIR) has deeper penetration depth (mm to cm), but the spectral absorbances are much weaker and not as well understood. Combining these two spectral regions may provide valuable information about the sample composition. Previous studies in the food industry have shown that combining NIR and MIR spectroscopy resulted in optimal differentiation of composition [2]. Mid-IR fiber optic probes have previously been used to differentiate between normal and pathologic connective tissues, and a recent study by our group has shown that the fiber optic probe spectral parameters correlate with cartilage histological grading [3, 4]. NIR fiber optic probes have been used during arthroscopy to evaluate the degree of degeneration of cartilage [5, 6]. The aim of this study was to combine and compare the use of MIR and NIR to differentiate regions within the ACL, and to differentiate ACL versus patellar tendon, as a preliminary study towards better understanding the ligamentization process in vivo. We hypothesize that the combination of NIR and MIR spectra will result in better differentiation compared to NIR or MIR spectroscopy alone. Methods: Bovine ACLs (n=3) and patellar tendons (n=3) were dissected from freshly slaughtered 2-14 days old calves (Green Village, NJ). NIR spectra were collected in diffuse reflectance mode using a 3 mm diameter NIR fiber optic probe (Art Photonics, Berlin, Germany) coupled to a Matrix-F infrared spectrometer (Bruker, MA). Spectra were collected from 2 points at the midsubstance, the femoral and tibial insertion sites of each ACL and patellar tendon (4000 to 11,000 cm -1 at 32 cm -1 spectral resolution with 128 co-added scans). At each data point 3 spectra were collected thus resulted in a total of 72 spectra. MIR spectra were collected from the same location as NIR data using a Thermo Scientific Nicolet is5 FT-IR spectrometer fitted with a fiber optic coupler (Harrick Scientific Products, Inc., Pleasantville, New York) and a silver halide attenuated total reflectance (ATR)-loop mid-infrared fiber optic probe (Art Photonics, Berlin, Germany) at 8 cm -1 spectral resolution, with 32 co-added scans in the frequency range of 600-2000 cm -1. Data processing: The spectra were processed using Unscrambler 10.1 (CAMO, NJ). The spectra were pretreated with a multiplicative scatter correction (MSC) followed by second derivative (Savitzky Golay, 3rd polynomial order, 11 point smoothing for MIR and 21 point smoothing for NIR data). A concatenated matrix was formed with NIR and MIR spectra where rows were comprised of NIR and MIR spectral absorbances from same sample as well as same location. MIR and NIR spectra were pretreated separately. Combined spectra were normalized by the standard deviation at each wavelength in the entire spectral collection. Separate partial least square discriminant analysis (PLS-DA) models with random cross validation were performed to differentiate ACL versus patellar tendon (11 segment with 6 samples) and insertion site versus midsubstance within the ACL (7 segments and 4 samples) using NIR spectra, MIR spectra and NIR and MIR combined together. Results: MIR spectra from ACL and patellar tendon were dominated by collagen peaks at 1650 (amide I), 1550 (amide II), 1338 (side chains) and 1240 cm -1 (amide III) which result from vibrations of the peptide bonds (also present to lesser amounts in proteoglycans, (PG)), and by PG sugar ring vibrations, 985-1140 cm -1, (Fig. 1A). NIR spectra of ACL and patellar tendon were dominated by water peaks at 5200 and 6890 cm -1 (Fig. 1B). To discriminate ACL and tendon, the best PLS-DA classification was based on MIR spectra alone, which resulted in 97.2 % accurate classification (Table 1). However, to discriminate insertion sites and midsubstance regions within the ACL tissue, PLS-DA based on combined use of NIR and MIR resulted in the best classification (87.1%, Table 1) Discussion: The loadings (which reflect the spectral features that contribute to the model) for the MIR spectra PLS-DA model of ACL versus patellar tendon classification were dominated by the amide II absorbance at ~1550 cm -1, likely reflecting differences in collagen and PG content in these two tissues at the surface. Ligament is a heterogeneous tissue, and its matrix composition
varies throughout the length and depth [7]. MIR alone did not perform well to classify different regions of ACL, likely due to the limited penetration of the MIR radiation which could not fully interrogate the ACL structure. However, addition of the NIR spectral region resulted in better discrimination between insertion sites and midsubstance within the ACL tissue. The loadings for PLS-DA model based on combined MIR and NIR spectral regions were dominated by water peaks at ~5200 cm -1 and 7000 cm - 1, and by matrix peaks at 1079 cm -1, 1250 cm -1, 1643 cm -1, 4300 cm -1 and 4700 cm -1. It should be noted that both spectral regions contributed towards differentiation of ACL regions, with the dominant frequencies arising from both water and matrix components. Significance: The combination of NIR and MIR spectral regions is most effective in assessing changes in ligament and tendon. This methodology could lead towards better understanding of healing of various orthopedic tissues, and effect of therapeutics and treatment modalities. Acknowledgments: This study was funded by NIH R01 AR056145 research grant. References: 1.Amiel, D., et al. J Orthop Res, 1986. 4(2): p. 162-72. 2.Dupuy, N., et al. Anal. chim. acta, 2010. 666(1): p. 23-31. 3.Hanifi, A., et al. Am J Sports Med, 2012 40(12):p.2853-61. 4.West, P.A., et al. Appl Spectrosc, 2004. 58(4): p. 376-81. 5.Spahn, G., et al. BMC Musculoskelet Disord, 2007. 8: p. 47. 6.Spahn, G., et al. Med Eng Phys, 2008. 30(3): p. 285-92. 7.Sabiston, P., et al. Clin Orthop Relat R, 1988. 236. 279-285
1988.
ORS 2014 Annual Meeting Poster No: 1392